Automated credit decisions
you can prove
Design and run your underwriting policy without code — with a trace behind every decision, a record behind every change, and an explanation for every applicant.
Check out our decisioning knowledge hub
Automated underwriting now has to be evidenced
Three regimes, one demand: show how the decision was reached, who changed the logic, and why this applicant got this answer.
DORA — applies since January 2025
Financial entities must manage ICT risk, document their arrangements and hold a workable exit strategy for every critical provider. Decisimo exports your decision logic in full, whenever you ask, and keeps a dated record of every change to it.
EU AI Act — creditworthiness is high-risk
Assessing the creditworthiness of a person is an Annex III high-risk use. The AI Omnibus moved those obligations to 2 December 2027: automatic logging, technical documentation and meaningful human oversight. The evidence takes longer to build than the policy does.
GDPR Article 22 — already today
A declined applicant can demand human intervention and a meaningful explanation of an automated decision. Decisimo replays any past case exactly as it ran, showing the data, the rules that fired and the scores behind the outcome.
Decisimo produces the traces, records and exports these regimes ask for. Compliance remains your responsibility — our job is to make it evidenceable.
Keep your credit policy current, and provable
Change the logic, test it against real cases, deploy it, and keep the evidence — without waiting for an engineering release.
Trace every decision
Every execution stores what came in, which data was pulled, which rules fired, what each model scored and how the outcome was reached. Open any past case and read it back.
Prove every change
Full revision history on your decision logic: who changed which rule, when, and what it looked like before. Export the evidence pack your auditor or regulator asks for.
Build the flow, not the release ticket
Drag-and-drop design for risk teams. Run champion and challenger side by side on live traffic, so a new policy proves itself before it becomes the policy.
Rules, tables, models — and LLMs
Business rules and decision tables for the policy, imported Python AI/ML models for the scoring, and LLM or agent calls with guardrails where judgement helps. All in one flow, all traced the same way.
Pull in the data the decision needs
Credit bureaus, blacklists, device and fraud signals, your own services — through any REST API, in real time, recorded against the decision that used them.
One API call from your core system
Integrate with a single REST call and deploy to regional endpoints, or to dedicated endpoints in your own region on Enterprise.
Ship a policy change today.
Prove it tomorrow.
